A Study of Designing and Development of Imagery Using Swarm Computing
Techniques
Thota Siva Ratna Sai1, Dr. Kunchala Suresh Babu2
1Research Scholar, Department of Computer Science and Engineering, Sri Satya Sai University of Technology &
Medical Sciences, Sehore, M.P., India.
2Research Supervisor Department of Computer Science and Engineering, Sri Satya Sai University of Technology &
Medical Sciences, Sehore, M.P., India.
Abstract
In recent years, the social bug metaphor for problem solving has been a hot subject, with an emphasis on stochastic construction practice, which focuses on constructing the key probabilistically to maximize the solution to any kind of problem. Swarms are a disorderly gathering / population of individuals moving in random directions and clustering together. Swarm Intelligence (SI) has been used to solve and optimize a variety of issues, including the Traveling Salesman Problem (TSP), Robotic Navigation, and Image Processing. Many techniques for processing images have been established, including conventional, statistical, knowledge-based, neural networks, and other Artificial Intelligence (AI) approaches; nevertheless, many academics are interested in investigating novel ways to improve accuracy. Existing techniques, however, are still limited as picture complexity grows, and they have many inherent drawbacks, such as: More processing time and effort are required. Imagery that has been processed is prone to mistakes. In terms of accuracy evaluation criteria such as Figure of Merit (FOM), kappa value, and Peak to Signal Noise Ratio, the result was inefficient (PSNR) To address these constraints, an automated image processing and optimization procedure may be developed, which should result in high speed and accuracy without compromising the quality of the input and output pictures.
Keywords: Imagery, Swarm Computing Techniques, Swarm Intelligence, Traveling Salesman Problem, Artificial Intelligence, Figure of Merit.
1. INTRODUCTION
The social representation of insects for addressing issues has become an emergent subject in the New Year that emphasizes stochastic development work and builds the key to updating the arrangements identified for any kind of problem. The organically driven world including the social experts emphasizes direct and aberrant connections by setting an example in the clear world by the pleasant conduct of specialists connecting locally with the climate. This dispersed variety/population of individuals moving in irregular ways, which in general are known as swarms. Swarm Intelligence (SI) is used to handle and simplify many problems, from a problem with the travelling salesman (TSP) to robotic navigation to image processing. Various imaging methods such as the customary, factual and information- driven neural organizations as well as other Artificial Intelligence (AI) methods have been developed; in any case, the investigation of new approaches to accuracy was key to the examination of numerous scientists.
In imaging science images are processed using numerical tasks or any type of signal preparing for which information is an image, such as a photograph or video contour; the yield may be an image or a set of qualities or limitations that are identified by the picture. The important issue in the handling of images is to find helpful data. It is becoming important in the New Year's welcome to experts from a few areas to offer processes that will properly deduce and handle an information picture and return age without limiting the quality of the information. Edge discovery and order are two key and emerging image processing components. Basically, edge locators subsequently significantly reduce the data measurement by scrutinizing less significant data while saving the fundamental underlying properties of an image. Order discovers useful information by utilizing the visual content of a picture and
organizing images in bundles based on their similarities and customer benefit. The interplay between image recovery and subsequent categorization may be performed in two different ways: text-based or content-based.
In the field of imaging science, image processing is performed using mathematical operations or any kind of signal processing when the input is a picture, such as a photograph or video frame; the output may be an image or a collection of image-related features or parameters. The key issue in the processing of images is the provision of constructive information. Researchers from many disciplines are becoming more important in recent years for proposing methods to effectively store and process a picture and output without sacrificing on quality. Edge detection and classification are two essential and developing picture processing components. Edge detectors essentially reduce object boundaries considerably by filtering out the less important information and retaining the key structural characteristics of a picture. Classification derives valuable information via the use of the visual content of an image and the classification of the pictures according to their similarities and interest. The picture retrieval and subsequent categorization procedure may be carried out in two ways: text-based or content-based.
The pictures are first evaluated on the basis of textual description in the text-based image recovery and then searched for the categorization. The major drawback of this method is to create descriptive texts manually and to continue searching in a big data base, so that a range of task-dependent 2 searches cannot be supported. The content- based image recovery uses colour, texture, form and spatial arrangement to represent and index the picture. The visual contents of the pictures are extracted by function vectors and stored in the function database. The user uses either the query picture or a designed image to feed the input. The system converts the input to its internal feature vector representation. The similarities between the query picture feature vectors and the image vectors in the feature database are then assessed such that the best match is obtained utilizing the indexing method. However, the growth in the complexity of a picture limits these techniques.
2. LITERATURE REVIEW
A.K.Deb (2012) this chapter presents an overview of several methods of 'soft computing' which try to imitate the inaccuracy and comprehension of natural events for the creation of algorithms. It details some of the prominent evolutionary computer methods including genetic algorithms (GA), PSO, ant colony optimization (ACO), and artificial immune systems (AIS). Introduce the concept of fuzzy sets and explain two techniques of inference, the Mamdani model and the Takagi–Sugeno–Kang (TSK) model. The origin and approximation of brain modelling in order to create neural networks that can learn are also addressed. There are two extremely common computer intelligence methods, vector support systems (SVMs) and rough sets. The concepts of hybridization that have inspired attention by utilizing the best characteristics of various methods in creating new algorithms are discussed.
Each section includes applications in different fields of the relevant method.
Maxwell. A. Asumeng (2015) over the last 20 years, models have been used to guide the process of organizational development consultancy (OD) for increased organizational effectiveness with consequences for the creation of learning organizations. This study analyses and synthesizes the features, similarities and differences and strengths and weaknesses of the four principal OD models and their application in the creation of learning organizations. The models examined include the three-stage, action research, appreciative research and the planned model of transformation. Whereas all four models overlap with features such as the participation of people in the change process, there are significant distinctions between them, including emphasis and phases of change. The study will revisit and expand the basic OD model that combines the other three models to solve two major shortcomings. The first is the lack of a stage in the process of change, which focuses on evaluating appropriate organizational and customer variables that may influence the success/failure of planned change initiatives.
Dheeraj Sundaragiri (2015) Image processing is a method for performing a particular operation on a picture, getting an updated image or separating some useful data from it. It is a kind of sign processing, in which the input is an image and the output may be image or attributes/highlights. Image processing is one of the fast-growing technologies nowadays. It organises the exploration centre area within the fields of design and software engineering.
Swarm techniques have a huge application area that includes different orders, including strength, fluffy framework, anticipation, bio-medicine, sociological research, picture handling, sound preparation, signal handling, data testing, measuring modelling, measurement managing, etc. Various methods and their variations have been established in the last 20 years. Whatever the number of variations, the basic skeleton of these processes remains the same. A significant part of these processes have been modified with various application areas to find a route to a particular
Khaleel, Shahbaa (2016) Due to the development in visual and sound innovation and its direct management in online media, it has attracted attention in light of its importance at present in the strategies of compacting pictures.
Because image pressure enables shading of picture data to be shown with the lowest amount of components, reducing transmission time in the organisation and increasing speed. The lossless pressure methods are used to ensure that the pressure cycle is completed without loss of data, since no data is lost while the pressure is being interacted. In this investigation a new framework has been developed to pack the shading pictures with higher productivity and grade.
Xiaohui, Ding and Huapeng (2017) Swarm intelligence algorithms have been widely used in the dimensional reduction of hyper spectral distant imaging detection. The ant colony algorithm (ACA), the clone determination algorithm (CSA) and the molecular swarm optimisation (PSO) are the main agent swarm intelligence algorithms used for selecting optimal band subsets as a sub-set age technique. Be that as it may, focus on their near-band appearance was rare. For this article, we used ACA, CSA, PSO, GA and a standard insatiable algorithm (particularly SFFS) as subset age systems and as objective capacities used the usual Jeffreys–Matusita distance (JM).
Kour, Vippon and Arora (2013) Pictures are the storage of visual information. They are handled, separated and described to extract the required data. As the data grows, assets are decreasing. The use of assets has to be monitored. The optimization methods have now replaced the existing conventional and ancient approaches in order to tackle this problem. These methods have been beneficial to mankind. We have taken these optimization techniques into account in our work. The Particle Swarm Optimization (PSo) and the Support Vector Machine classifier have been thoroughly examined for optimization methods (SVM).
Rostami, Mehrdad and Berahmand (2015) in the previous many years, the fast development of PC and database advances has prompted the quick development of enormous scope datasets. Then again, data mining applications with high dimensional datasets that require fast and exactness are quickly expanding. An important issue with these applications is the scourge of dimensionality, where the quantity of highlights is a lot higher than the quantity of examples. One of the dimensionality decrease approaches is highlight choice that can build the exactness of the data mining task and lessen its computational intricacy. The element determination strategy targets choosing a subset of highlights with the most minimal internal closeness and most elevated importance to the objective class.
Nagarajan, Bharanidharan (2012) numerous delicate figuring strategies are utilized these days to investigate clinical images, and determination of infection is automated. This paper looks at the presentation of Weighted Artificial Bee Colony and Particle Swarm Optimization in the finding of dementia utilizing MRI images. For examination, cross- sectional MRI of 235 subjects gathered from OASIS is utilized. By changing the loads for both optimization strategies in a legitimate way, streamlined outcomes can be reached. These procedures arrange the cross-sectional image into three classifications and give practically equivalent Goodness Detection Ratio of 78% alongside various relapse proportions.
Tabib Mahmoudi (2017) Object-based image analysis (OBIA) methods are often superior than pixel-based pictures discovered remotely by Very High Resolution (VHR). Due to the comparative markings of other-world land cover classes in urban areas in particular, geographical data should be exploited to build superior classification maps for better targets. Characterization based on division and rules are the main two phases of the powerful OBIA method, which has a wide variety of use in design recognition and design applications. The selection of the finest characteristics for divisional borders has a major impact on the division outcomes. When picture objects are identified, topological relationships between them, fact-related highlights and form highlights from other worlds or textures may all be used for standard-based characterisation.
3. SWARM INTELLIGENCE (SI)
SI is an artificial intelligence technique based on the study of collective behavior in distributed and self-managed system, where the artificial agents interact among themselves and also with the environment locally leading to some global pattern to emerge. Although there is normally no centralized controlling or managing structure dictating how individual agents should perform, local communications between such agents often lead to the realization of global behavior. Many such examples can be found in nature including ant colonies, bird flocking, animal herding, bacteria molding and fish schooling.
1. Computational Swarm Intelligence
An optimal solution for combinatorial problems, which are NP-Hard in nature, is difficult to achieve using traditional algorithms as well as evolutionary algorithm. Hence, Studies of the social behavior of individuals in swarms prompted to design very efficient optimization and clustering algorithms. The self-organizing principle of Swarm Intelligence have been successfully applied in a variety of problem domains including function optimization problems, finding optimal routes, scheduling, structural optimization, and image and data analysis. Based on Swarm Intelligence concept various meta-heuristics are being developed which are generally termed as SI Models. Ant Colony Optimization (ACO), Particle Swarm Optimization(PSO), Artificial Bee Colony (ABC) optimization, Cockroach Swarm Optimization (CSO) and African Buffalo Optimization are some of the models which have shown the promising behavior and widely used in the field of optimization. The next section describes the Swarm Intelligence models and their computational significance.
Ant Colony Optimization (ACO)
Ant Colony Optimization (ACO) based algorithms intend to the study of computational systems in which computation is carried out by artificial ants to imitate the behaviour of read ants. As shown in fig. 1, their fundamental principle is based on the way in which ants searching for their food and returning back to the nest. Real ants are used to find food source without any visual information and this capability of real ants are also used in finding shortest path between nests to food source. Initially, ants search for the food in random manner and explore the surrounding area of the nest. Whenever ant finds a food source it evaluates the quantity and quality of food and carries the required food return to the nest. During the return journey, ant leaves a chemical substance known as pheromone trail on the ground. The other ants follow this pheromone trail to find a food source for them. After some time, pheromone intensity will be increased at one particular path and all the ants start following that path only.
Hence, ants exploit these all pheromone trails as a medium of finding their way from source to destination and back.
Figure 1 Real ant’s social behavior
ACO Met heuristics
Ant colony optimization is a metaheuristic for difficult combinatorial optimization problems modelled after the stigmergetic communication of ants finding shortest paths to food sources. The first ACO-algorithm was Ant System (AS), introduced by Dorigo in 1992. He later generalized it into the ACO metaheuristic. A metaheuristic is a set of algorithmic concepts that can be used to define heuristic methods application to a wide set of different problems.
4. PARTICLE SWARM OPTIMIZATION (PSO)
Particle Swarm Optimization based on birds flocking behaviour. Bird flocking can be defined as the social collective motion behaviour of a large number of interacting birds with a common group objective. The local interactions among birds (particles) usually emerge the shared motion direction of the swarm. Such interactions are based on the
―nearest neighbour principle where birds follow certain flocking rules to adjust their motion (i.e., position and velocity) based only on their nearest neighbours, without any central coordination.
1. PSO Met heuristics
Particle Swarm Optimization (PSO) is another heuristic optimization technique of swarm intelligence model, which was introduced by Russell Eberhart and James Kennedy in 1995 (2). PSO is a population-based search strategy that finds optimal solutions using a set of flying particles with velocities that are dynamically adjusted according to their historical performance, as well as their neighbours in the search space. PSO solves problems whose solutions can be represented as a set of points in an n-dimensional solution space. The term ―particles refers to population members, which are fundamentally described as the swarm positions in the n-dimensional solution space. Each particle is set into motion through the solution space with a velocity vector representing the particle‘s speed in each dimension.
Each particle has a memory to store its historically best solution.
5. SWARM INTELLIGENCE ALGORITHMS
Swarm Intelligence (SI) is an artificial intelligence discipline, was introduced by Gerado Beni and Jing Wang in 1989. A swarm is considered to be an arranged collection of interacting creatures into a system. Computational study of swarm intelligence includes the algorithms of ants, bees, wasps, termites, fish in schools and birds in flocks.
PSO based Scheduling and Load Balancing Algorithms
The PSO algorithm incorporates local search methods with global search methods for trying to equalize exploration and exploitation. PSO is a popular technique due to its functionality in large range of applications with little computational cost and its simplicity. It is also a kind of meta-heuristic technique. The major objectives of PSO are:
balance load of the system, minimize time and communication cost and increase scalability. The authors discussed in, proposed PSO to overcome the main challenge of cloud computing. This challenge is load balancing because nowadays the number of users in different organizations using cloud computing are increasing for their requirements so the performance of cloud computing is decreasing. This paper focused on minimizing makespan time. The authors in proposed their scheduling method based on improved PSO (IPSO) for workflow applications and mainly focused on workflow scheduling in cloud-fog environment to gain the optimal solution such as more cost-effective and better performance than PSO approach. Rodriguez, et al. proposed resource provisioning and scheduling using PSO for minimizing the execution cost of the whole works passes through form the beginning until it is finished and make span as deadline constraints.
ABC based Scheduling and Load Balancing Algorithms
The ABC algorithm is one of the most newly optimization. It is also a swarm based meta-heuristic algorithm. The good points of this algorithm include: simplicity, flexibility, robustness, need fewer control parameters compared to others searching techniques, and minimize cost with stochastic nature. The authors in proposed new optimization method called Bees Life Algorithm (BLA) to solve the issue of numerous computational resources requested by mobile users and to ensure the efficient execution of tasks in the fog computing environment. The job scheduling algorithm fixed an optimal task of various jobs submitted to be executing on tradeoff between CPU execution time and memory required by fog computing, using for infrastructure. This Algorithm resulted show better performance of CPU execution time and allocated memory than GA and PSO algorithms. But, this paper needed to consider dynamic job scheduling for the arrival of new requests are being executed in fog computing environment. The authors in proposed an effective Improved Honey Bee Algorithm in cloud computing to be better make span for both dynamic and static task scheduling.
6. IMAGE PROCESSING
Image Processing considers the image as a two-dimensional array of intensities of each pixel as an input and applies several mathematical operations to it. The various operations performed in Image Processing includes Image Acquisition, Image Enhancement, Image Restoration, Image Compression, Image Segmentation, 4 Representation &
Description, Object Recognition and Image Classification The proposed work aims to highlight the improved techniques for processing Imagery using Swarm based methods for Edge Detection, Feature Extraction, Feature Selection and Image Classification.
Edge Detection
Edge is an important feature of an image which carries key information about the objects present in the image.
Depiction of edges is known as edge detection which is by far the most familiar advancement made for detecting meaningful discontinuities in intensity values. It significantly lessens the image content to be processed for the high- level processing assignments such as object recognition, image segmentation etc. The success of production of true edge map depends on the determination of threshold value. Various edge detection methods available in the literature includes Roberts, Sobel, Prewitt edge detectors, LoG edge detector, MarrHildreth edge detector, Canny edge detector, Basic declivity edge detector, Local threshold and Boolean function based edge detector etc.
7. IMAGE CLASSIFICATION
It is always preferable to have an image, showing a magnitude of colors illustrating various features of the underlying terrain, but it is useless unless we know what the colors in an image signify. The key steps in image classification are definition of classes selection of features, sampling of training data, estimation of universal statistics, application of appropriate classifier and verification of the results. The classified results should be checked and verified for the accuracy with reference to the real data. Major applications of image classification are biomedical imaging, biometry, video surveillance, vehicle navigation, industrial visual inspection, robot navigation, and remote sensing etc. Maximum-Likelihood Classifier (MLC) is the most commonly and widely used technique which determines the likelihood between the known and unknown pixels in the sample data. The specification of the sample data is the key to the overall success of classification method. The other method viz; Knowledge Based System classifier (KBS) though easy to understand, is constrained by the long and repetitive process of acquiring information from the experience which may be biased. Neural network based classifier is best suited for parallel estimation but the rules derived from it are difficult to infer. Artificial Intelligence and its theory thus provide a scope for image classification.
8. CONCLUSIONS
The use of social insect behavior to solve different issues has become an important field of study in order to develop and improve solutions for any kind of real-time challenge. The biologically inspired world of social agents stresses interaction between agents and with the environment, resulting in the development of some global pattern that serves as a solution to different issues. Swarm intelligence (SI) is a branch of artificial intelligence that studies the collective behavior of swarms and has been used to a variety of fields, including image processing. Without depending on written communications, images allow humans to perceive natural things. Many techniques for processing images have been established, including conventional, statistical, knowledge-based, neural networks, and other artificial intelligence approaches; nevertheless, many academics are interested in researching novel ways to improve accuracy. 121 Various approaches utilizing swarm computing techniques, such as ACO, PSO, and FPAB, have been devised and developed for edge detection and classification in the suggested study effort. Module-I (AASC) for edge detection and Module-II (IASC-CI), Module-III (IAPSO-TCI), and Module-IV (IAABO-TCI) for classification have been suggested. Pictures of different sizes may be used to test the suggested modules, however for the above comparison, images of the same size, 512 X 512, are used. The accuracy evaluation metrics such as FOM, Kappa value, PSNR value, overall accuracy, and overall kappa are unaffected by picture size. However, when the size of the picture changes, the execution time rises or reduces.
9. REFERENCE
1. A.K.Deb 2012,” Introduction to soft computing techniques: artificial neural networks, fuzzy logic and genetic algorithms,” Soft Computing in Textile Engineering Wood head Publishing Series in Textiles 2011, Pages 3- 24
2. Maxwell. A. Asumeng 2015,” Organization Development Models: A Critical Review and Implications for Creating Learning Organizations,” European Journal of Training and Development Studies Vol.2, No.3, pp.29-43, September 2015
3. Dheeraj Sundaragiri 2015,” A Study of Classification of Different Imagery Using Swarm Computing Techniques,” Research Supervisor Department of Computer Science Engineering, Sri Satya Sai University of Technology & Medical Sciences, Sehore, M.P., India.
4. Khaleel, Shahbaa. (2016). Image Compression Using Swarm Intelligence. International Journal of Intelligent Engineering and Systems. 14. 257-269. 10.22266/ijies2021.0228.25.
5. 2. Xiaohui, Ding & Huapeng, Li & Yong, Li & Ji, Yang & Shuqing, Zhang. (2017). Comparison of swarm intelligence algorithms for optimized band selection of hyperspectral remote sensing image. Open Geosciences. 12. 425-442. 10.1515/geo-2017-0155.
6. Kour, Vippon & Arora, Sakshi. (2014). Vision based Techniques for Image Classification: A Survey.
10.2139/ssrn.3562965.
7. Rostami, Mehrdad & Berahmand, Kamal & Forouzandeh, Saman. (2015). Review of Swarm Intelligencebased Feature Selection Methods. 5. Qi, Yingji & Ding, Feng & Xu, Fangzhou & Yang, Jimin.
(2013). Channel and Feature Selection for a Motor Imagery-Based BCI System Using Multilevel Particle Swarm Optimization. Computational Intelligence and Neuroscience. 2015. 1-11. 10.1155/2015/8890477.
8. 6. Nagarajan, Bharanidharan & Rajaguru, Harikumar. (2012). Comparison of Particle Swarm Optimization and Weighted Artificial Bee Colony Techniques in Classification of Dementia Using MRI Images.
10.1007/978- 3-030-00665-5_95.
9. 7. Mohsin, Ali & Zaidan, A. & Bahaa, Bilal & Albahri, O.s & Albahri, A.s & Alsalem, Mohammed &
Mohammed, K. & Enaizan, & Ameen, Hussein & Garfan, Salem. (2014). New Method of Image Steganography Based on Particle Swarm Optimisation Algorithm in Spatial Domain for High Embedding Capacity. IEEE Access. PP. 1-1. 10.1109/ACCESS.2014.2949622.
10. 8. Priya, T. & Palanisamy, Kalavathi. (2012). Brain Tissue Segmentation in MRI Brain Images using Histogram based Particle Swarm Optimization Techniques. Current Medical Imaging Formerly:Current Medical Imaging Review. 15. 10.2174/1573405615666190318154943.
11. 9. Tabib Mahmoudi, Fatemeh. (2015). Swarm Intelligence for Object Based Image Analysis.
10.1049/PBCE119H_ch7.
12. 10. Zajmi, Leke & Ahmed, Falah & Jaharadak, Adam Amril. (2016). Concepts, Methods, and Performances of Particle Swarm Optimization, Backpropagation, and Neural Networks. Applied Computational Intelligence and Soft Computing. 2010. 1-7. 10.1155/2010/9547212.
13. Snehlata Sheoran 2009,” Analysis on application of swarm-based techniques in processing remote sensed data,” Earth Science Informatics volume 13, pages97–113 (2009)Cite this articl 204 Accesses 2 Citations Metrics
14. Ruiz-vanoye J a, Díaz-parra O, Cocón F, Soto A, Ángeles MDL, Arias B, et al. Meta-Heuristics Algorithms based on the Grouping of Animals by Social Behavior for the Traveling Salesman Problem. International Journal of Combinatorial Optimization Problems and Informatics. 2012;3(3):104–123.
15. Zhu S, Li G, Han J. An improved PSO algorithm with object- oriented performance database for flight trajectory optimization. Journal of Computers. 2012;7(7):1555–1563.
